Observe

Start with the Dashboard for a fleet summary, Apps for connection health, and Agents for individual status. The Live Monitor brings operational events and command activity into one view. Open a run or command when you need the evidence behind an alert.

Two upward calls. The heartbeat carries current status and counters since the last beat; the run report carries one completed run. Both are wrapped so a failure only logs.

The Runs screen collects execution records from the integration and from runs started in the panel. Filter by app, agent, status, and time, then open the input, output, model, token counts, and latency that were actually reported.

The Tokens screen shows input and output usage over the selected date range. Use the agent, app, and model breakdowns to locate the source of a change. Cost is an estimate based on reported usage and the model pricing available to ACP, not a provider invoice.